1st, I said that the current could be controlled using a resistor that was chosen so the voltage drop at the desired current equaled about 1 volt and then use a pot to feed the Q1 transistor, haven't made that work yet at higher currents!
2nd Refer back to the drawing in post #32, Q2 current limiting transistor on pin 4 of the timer, built that and the 40 volts came up and the 900ma current only came up to about 500 to 700 ma depending on which build I was running, then after a short time the current would drop and lights would dim. Some times, they would pulse on and off are a rate of about one or two hz, and with these LEDS, that's lots of very bright light flashing. Tried various fixes and nothing worked. Finally, I decided to try running the circuit as a voltage device instead of a current device and used a pot to control the voltage without any load, worked great, set the voltage to what I knew the LEDs could handle and then hooked up the LEDs while it was running, the lights came on, the current came up and all was good, for about 5 seconds, then the current started dropping and lights went dim. After several repeats and watching the scope as I did so, I noticed Vrst would be at ~2 and then drop to near 0 after I hooked the lights up, finally pulled Q2 and all was happy, I could set the voltage where I wanted so the current was what I wanted and every thing was basically stable except for changes as things heated up. The components did not get warm enough to be of concern, just the LEDs.
I think the design is good, but the oscillating of the inductor caused enough spikes to keep turning on Q2 which slowly draind C4 to the point that Vrst was too low to allow the timer to operate.
More playing tomorrow. Pretty neat, I was getting a full 40 watts out of it, guess I'll have to see how much I can get once I get the Q2 problem ironed out.
Kinarfi